In a groundbreaking move, the World Health Organization (WHO) has released comprehensive guidance on the ethics and governance surrounding large multi-modal models (LMMs), a rapidly evolving segment of generative artificial intelligence (AI) with profound applications in healthcare.
Understanding LMMs: A Revolution in AI Technology
LMMs, characterized by their ability to process various data inputs like text, videos, and images, have ushered in a new era of AI. With capabilities resembling human communication, they can perform tasks beyond explicit programming. Platforms like ChatGPT, Bard, and Bert gained widespread recognition in 2023, making LMMs an integral part of our technological landscape.
WHO's Call to Ethical Action
The WHO's guidance encompasses over 40 recommendations directed at governments, tech entities, and healthcare providers. Dr. Jeremy Farrar, WHO Chief Scientist, emphasizes the potential of generative AI in healthcare but underscores the necessity of addressing associated risks. Transparent information and policies, according to Dr. Farrar, are pivotal in harnessing LMMs for better health outcomes.
Applications and Risks of LMMs in Healthcare
The guidance outlines five broad applications of LMMs in health, ranging from clinical care and patient-guided use to education and scientific research. Despite their potential benefits, LMMs pose documented risks, including the generation of false or biased information. The quality and bias of training data raise concerns about the accuracy of health-related outputs.
Addressing Challenges and Ensuring Ethical AI
The WHO guidance highlights risks such as 'automation bias' and cybersecurity threats, emphasizing the need for collaboration among stakeholders. Governments, technology companies, healthcare providers, patients, and civil society are urged to engage in the development and deployment of LMMs, ensuring ethical oversight and regulation.
Key Recommendations for Governments and Developers
Governments are called upon to invest in public infrastructure, enact laws for ethical obligations, and establish regulatory agencies for LMM assessments. Mandatory post-release auditing and impact assessments by independent third parties are proposed to ensure accountability.
Developers are urged to engage diverse stakeholders in transparent design, incorporating ethical considerations. LMMs should be designed for well-defined tasks with accuracy, reliability, and predictability. The involvement of potential users, medical providers, and patients from the early stages is emphasized.
Paving the Way for Ethical AI in Healthcare
As the WHO spearheads ethical guidelines for LMMs, the world is on the cusp of a transformative journey in healthcare AI. Striking a balance between innovation and ethical considerations is essential to harness the full potential of LMMs, ensuring they contribute positively to global health outcomes. (Reference: https://www.who.int/news/item/18-01-2024-who-releases-ai-ethics-and-governance-guidance-for-large-multi-modal-models)